RESEARCH PROJECTS

Book cover imagePSIMAPP: Progressively Segmented Image Modeling As Poly-Patterns

PSIMAPP does modeling of images as patterns of pixels and properties to produce Image Models As Grouped Elements (IMAGE). The grouped elements can be picture elements (pixels), groups of pixel groups, groups of grouped groups, or grid groups. The basic modeling is accomplished by doubly segmenting the images to obtain poly-pattern models as approximating mosaics of segments at two levels of detail.

The more coarsely aggregated A-level consists of 250 segments that are numbered in order of overall intensity and recorded as a byte binary map of segment numbers in a file having a .BSQ extension.

The finer B-level of base segments has several thousand pattern partitions that are nested within A-level segments up to a maximum of 255 B parts in an A part. A second byte binary file having a .BIL extension serves as a binary index layer for mapping base segments. Several supplemental files contain other tabulations that support the pattern mappings. The patterns of the B-level base segments provide for approximate (smoothed) restoration, and the composite patterns of A-level provide for rendering of pattern pictures having controlled contrast, conducting contextual and content analysis, compiling change, generating generalizations and much more. There is a Windows-style "front end" for the more frequently used modules. All modules can be applied via a self-documenting .RUN file of textual specifications.

Download PSIMAPP program files
View PSIMAPP overview presentation (powerpoint file)
View abstract: Progressively Segmented Image Modeling as Poly-Patterns (PSIMAPP) for Landscape Analysis
Read description of book (Book available from Springer)